An Empiric Risk Score to Guide Presumptive Treatment of Asymptomatic Anorectal Infections in Men Who Have Sex With Men in Kisumu, Kenya

Abstract

Introduction The World Health Organization (WHO) recommends presumptive therapy (PT) for Neisseria gonorrhoeae (GC) and Chlamydia trachomatis (CT) in asymptomatic MSM who report unprotected receptive anal intercourse in the past 6 months and either multiple sex partners or a partner with an STI. We aimed to identify predictors of asymptomatic anorectal infection in Kenyan MSM and compare performance of an empiric, model-based risk score to that of the WHO PT algorithm. Methods Anorectal GC/CT infections were diagnosed at baseline among 698 MSM enrolled in the Anza Mapema study in Kisumu, Kenya. Study clinicians conducted and recorded a standardized medical history and physical examination findings (including genital and rectal examination). Rectal swabs for GC and CT screening were obtained by proctoscopy or self-collection. The Abbott RealTime GC/CT NAAT was used to diagnose anorectal infections. Multivariable logistic regression was used to identify associations with asymptomatic anorectal GC/CT infection, after excluding participants with anorectal symptoms. We derived a total risk score (range: 0-5) for each participant using the coefficients of the final multivariable model. Risk score algorithm performance was compared to WHO algorithm performance with respect to sensitivity, specificity, and number needed to treat (NNT). Results Asymptomatic anorectal GC/CT infection prevalence was 4.2%. Predictors and corresponding risk scores were: HIV infection (2), age 18-24 years (2), and unprotected anal sex (1). A risk score ≥3 was 83% sensitive and 65% specific in detecting asymptomatic GC/CT anorectal infection. In contrast, the WHO PT algorithm had low sensitivity (25%), but was 84% specific. While 37% of asymptomatic participants met PT eligibility criteria using a risk score ≥3, only 17% met eligibility by WHO PT criteria. Using our risk score algorithm, 12 participants would need PT to treat one anorectal GC/CT infection, compared to 38 participants by WHO criteria. Conclusion An empiric risk score based on age, HIV status, and unprotected anal sex improved both sensitivity and efficiency (i.e., NNT) of identification of asymptomatic anorectal GC/CT infection, compared to the WHO PT algorithm. If validated in other settings, this risk score could improve the management of asymptomatic anorectal GC/CT infections in settings where diagnostic testing is not available.